{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/scisports-learning-football-kinematics","title":"SciSports: Learning football kinematics through two-dimensional tracking data","arxiv_id":"1808.04550","date":"2018-08-14","proceeding":null,"authors":["Anatoliy Babic","Harshit Bansal","Gianluca Finocchio","Julian Golak","Mark Peletier","Jim Portegies","Clara Stegehuis","Anuj Tyagi","Roland Vincze","William Weimin Yoo"],"abstract":"SciSports is a Dutch startup company specializing in football analytics. This\npaper describes a joint research effort with SciSports, during the Study Group\nMathematics with Industry 2018 at Eindhoven, the Netherlands. The main\nchallenge that we addressed was to automatically process empirical football\nplayers' trajectories, in order to extract useful information from them. The\ndata provided to us was two-dimensional positional data during entire matches.\nWe developed methods based on Newtonian mechanics and the Kalman filter,\nGenerative Adversarial Nets and Variational Autoencoders. In addition, we\ntrained a discriminator network to recognize and discern different movement\npatterns of players. The Kalman-filter approach yields an interpretable model,\nin which a small number of player-dependent parameters can be fit; in theory\nthis could be used to distinguish among players. The\nGenerative-Adversarial-Nets approach appears promising in theory, and some\ninitial tests showed an improvement with respect to the baseline, but the\nlimits in time and computational power meant that we could not fully explore\nit. We also trained a Discriminator network to distinguish between two players\nbased on their trajectories; after training, the network managed to distinguish\nbetween some pairs of players, but not between others. After training, the\nVariational Autoencoders generated trajectories that are difficult to\ndistinguish, visually, from the data. These experiments provide an indication\nthat deep generative models can learn the underlying structure and statistics\nof football players' trajectories. This can serve as a starting point for\ndetermining player qualities based on such trajectory data.","url_abs":"http://arxiv.org/abs/1808.04550v1","url_pdf":"http://arxiv.org/pdf/1808.04550v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"scisports-learning-football-kinematics","repo_url":"https://bitbucket.org/AnatoliyBabic/swi-scisports-2018","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"two","task_name":"Vocal Bursts Valence Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}